Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/269768970

ExpandedTemporalBindingWindowsinPeoplewithMildCognitiveImpairment

ARTICLEinCURRENTALZHEIMERRESEARCH·DECEMBER2014

ImpactFactor:3.8·DOI:10.2174/1567205012666141218124744·Source:PubMed

DOWNLOADS

25

VIEWS

70

7AUTHORS,INCLUDING:

JochenKaiser

Goethe-UniversitätFrankfurtamMain

133PUBLICATIONS4,042CITATIONS

SEEPROFILE

DavidPrvulovic

Goethe-UniversitätFrankfurtamMain

86PUBLICATIONS2,525CITATIONS

SEEPROFILE

MichaelJHogan

NationalUniversityofIreland,Galway

102PUBLICATIONS502CITATIONS

SEEPROFILE

MarcusJNaumer

Goethe-UniversitätFrankfurtamMain

35PUBLICATIONS801CITATIONS

SEEPROFILE

Availablefrom:JasonSeehoChan

Retrievedon:10July2015

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

1

Expanded temporal binding windows in people with mild cognitive impairment

Jason S. Chan1*, Jochen Kaiser1, Mareike Brandl1, Silke Matura2,3, David Prvulovic2,

Michael J. Hogan4, Marcus J. Naumer1

1Institute of Medical Psychology, Goethe-University, Frankfurt am Main, Germany 2Department of Psychiatry, Psychosomatics and Psychotherapy, Goethe-University,

Frankfurt am Main, Germany 3Institute of General Practice, Goethe-University, Frankfurt am Main, Germany

4School of Psychology, National University of Ireland, Galway, Ireland

SUBMITTED TO: Current Alzheimer Research

*Corresponding Author

Jason S. Chan chan@med.uni-frankfurt.de

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

2

Abstract

Previous studies investigating mild cognitive impairment (MCI) have focused

primarily on cognitive, memory, attention, and executive function deficits. There has

been relatively little research on the perceptual deficits people with MCI may exhibit.

This is surprising given that it has been suggested that sensory and cognitive functions

share a common cortical framework [1]. In the following study, we presented the

sound-induced flash illusion (SiFi) to a group of participants with mild cognitive

impairment (MCI) and healthy controls (HC). The SiFi is an audio-visual illusion

whereby two-beeps and one-flash are presented. Participants tend to perceive two

flashes when the time-interval between the auditory beeps is small [2; 3]. Participants

with MCI perceived significantly more illusions compared to HC over longer auditory

time-intervals. This suggests that MCIs integrate more (arguably irrelevant)

audiovisual information compared to HCs. By incorporating perceptual tasks into a

clinical diagnosis it may be possible to gain a more comprehensive understanding into

the disease, as well as provide a more accurate diagnose to those who may have a

language impairment.

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

3

Introduction

There is a great deal of controversy regarding the predictive validity of

standard neuropsychological assessments in identifying early cases of dementia [4; 5]

and the underlying causal factors associated with age- and disease-related decrements

in cognition [6-9]. Theoretical models of neurological aging have proposed a variety

of markers of generalized cognitive decline, including increased intra-individual

variability in “intellectual function” [10; 11], reduced processing speed [12], and

sensori-motor function decline [13]. It has recently been suggested [14] that research

on multisensory integration is warranted as it may facilitate the identification of more

sensitive markers of age- and disease-related cognitive decline and also enhance the

potential for early intervention and therapeutic applications, for example, focusing on

multisensory stimulation [15]. This is consistent with recent efforts to identify novel

behavioural and biological markers of mild cognitive impairment (MCI) that can be

used to inform choice of therapy and aid a personalized approach to clinical treatment

[16; 17].

Mild cognitive impairment (MCI) is an intermediate state between normal

aging and Alzheimer’s disease. It is characterized by impaired memory performance

or general intellectual function, more severe than normal aging, but not as severe as

compared to patients with Alzheimer’s disease [18-21]. The annual rate of

progression from MCI to dementia is 5% - 10%. Thecumulativerateisabout30%‐

40%.Thus,30‐40%ofpeoplewithMCIeventuallyconverttodementia. This is

substantially higher than the 1-2% for healthy older adults [22]. The incidence rates

of MCI vary widely, depending on how this syndrome is categorized. However, best

estimates are that the incidence of amnestic MCI subtypes range between 9.9 and 40.6

per 1,000 “person-years”, and the incidence of non-amnestic MCI subtypes are 28 and

36.3 per 1,000 person-years [18; 20; 23].

Previous studies investigating MCI have focused primarily on cognitive [24;

25], memory [26], attention [27], and executive function deficits [28]. Additionally,

people with MCI also have perceptual deficits. Previous research has provided

evidence that people with MCI have difficulty in speech perception, a dichotic digit

test [29; 30], as well as impaired visual motion perception [31]. However, to date,

there is little research which has investigated the multisensory integration of older

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

4

adults with MCI. One such study conducted by Wu, Yang, Yu, Li, Nakamura, Shen,

et al. [32] investigated the audiovisual integration using a simple reaction time

paradigm. They asked healthy controls (HC), MCI participants, and participants with

Alzheimer’s disease (AD) to indicate as quickly as possible when an attended

stimulus was presented to the right or left of the fixation cross. They found neither

significant differences in accuracy nor reaction times between HC and MCI.

However, when examining the cumulative distribution function, they found that MCI

participants violated the race model at later reaction times compared to HC. The race

model is a reaction time measure of multisensory integration. A violation of the race

model suggests that the effect is not due to a simple linear summation of the visual

and auditory stimuli.

Healthy older adults tend to integrate more audio-visual information over a

‘wider’ temporal binding window (TBW) and spatial area compared to young adults

[33-35]. The TBW is a span of time whereby information from the different sensory

modalities is combined to form a single percept of an object. In an audio-visual

temporal order judgement task, Setti et al. [36] found that young adults are able to

accurately determine which stimulus was presented first (auditory or visual) when the

stimulus onset asynchrony (SOA) was 70 ms or greater. However, older adults

required an SOA greater than 270 ms to accurately determine which stimulus was

presented first. This means that within the time span of 0 ms – 270 ms older adults

were combining the auditory and visual stimuli and believed they were presented

simultaneously. Using EEG they found that older adults exhibit smaller posterior P1

and fronto-central N1 components compared to the young adults. They suggested that

these reduced components reflect a deficit in cross-sensory processing in the older

adults.

Several other studies have used the sound-induced flash illusion (SiFi) to

investigate the temporal dynamics of multisensory integration in older adults [33; 37].

The sound-induced flash illusion (SiFi) is an audio-visual illusion where a single flash

is presented, along with two auditory beeps. Participants tend to perceive two flashes,

if the two beeps are presented within a short temporal interval [e.g., 70 ms; 2; 3]. For

young adults, if the SOA between the two beeps exceeds approximately 150 ms they

will perceive one flash. Older adults will continue to perceive the SiFi at longer

SOAs [33; 37]. Furthermore, older adults who are prone to falling perceived more

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

5

illusions at longer SOAs, compared to healthy controls [33]. The “fallers” had

normal unimodal visual and auditory perception. Setti et al. (2011) hypothesized that

although the SiFi is unrelated to balance (participants performed the task while

seated), the results indicate that the “fallers” overall have a reduced level of cortical

connectivity between the sensory areas.

The SiFi is considered to be an early sensory illusion. EEG studies have

demonstrated that the perceived illusion enhances early ERP components (starting

~30 ms) in the occipital lobe [38]. Using MEG, Keil, Müller, Hartmann, Weisz [39]

found that the perception of the SiFi was modulated by increased beta-band activity

(12 Hz - 21Hz) in the left middle frontal gyrus before the stimuli were presented.

Using phase-locking coherence, Keil et al., found increased phase coherence between

BA39, BA21, and BA18 in the beta-band for the perceived illusion trials. Beta-band

activity is considered to be a top-down signal, sending information from higher-level

areas to the early-sensory areas [40]. Keil et al., suggested that this increase in pre-

stimulus beta-band activity shows an enhanced predisposition to integrate audio-

visual information which is modulated at an inter-trial basis.

It is possible that the increased number of perceived SiFi, demonstrated in

previous studies, is due to a difference in focused attention between older and younger

adults. Older adults may have more difficulty in attending to a visual stimulus when

an auditory stimulus is presented, thus their attention may switch to the auditory

modality. DeLoss et al. [41] correlated performance in an auditory-only and visual-

only Go/NoGo task to the number of perceived SiFi. They found that, in accordance

to previous literature, older adults perceive more illusions. However, they found no

difference between the age groups in the Go/NoGo tasks; suggesting that the increase

in perceived illusions is not a product of increased distractor susceptibility or

executive function [37].

While several studies have suggested that healthy older adults integrate more

information of the different sensory modalities compared to younger adults; some

research has suggested that age related cognitive disorder (such as: Alzheimer’s

disease) is related to a “connectivity breakdown” due to a “loss of structural and

functional integrity of long cortico-cortical tracts” [42; see 43 for a review]. This

breakdown can result in a decrease in multisensory integration. Furthermore, the

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

6

superior temporal sulcus, one area responsible for multisensory integration is directly

affected by Alzheimer’s disease [50% loss of volume compared to healthy controls;

44]. More generally, it has been proposed that the dual role played by frontal and

cerebellar degeneration and the disruption of fronto-cerebellar feed-back and feed-

forward control loops may be of central importance for understanding age-related

changes in: multisensory integration, the timing of information processing, processing

speed, intra-individual variability of reaction times, and automaticity of cognitive

functions [6; 45]. These losses in connectivity between brain regions as well as

neuronal loss in the heteromodal areas can result in declined perceptual integration.

These neuronal losses can have a significant behavioural impact. Delbeuck et al. [46]

demonstrated that people with AD perceive fewer audiovisual McGurk illusions

compared to the HCs, thus providing behavioural evidence for these neuronal losses.

The goal of the present study was to determine if older adults with MCI also

integrate more audio-visual information, compared to healthy controls. We presented

the SiFi to people with MCI and age and gender matched healthy controls (HC), with

variable stimulus-onset asynchronies (SOAs) between the auditory stimuli. It is

possible that MCI participants perceive fewer illusions compared to HC because of a

decreased ability to integrate information from the auditory and visual modalities.

Conversely, it has been demonstrated that the multisensory processing in MCI is more

similar to HC than AD. Golob, Miranda, Johnson, and Starr [47] explored

crossmodal interactions using EEG in healthy older adults, older adults with MCI, and

those with AD. Golob et al. exploited a neuronal effect called the refractory period,

whereby there is a reduction in a component’s amplitude or latency due to the

presentation of a previous stimulus [48]. When two stimuli were presented in a single

modality (i.e., visual- or auditory-only), there was a reduction in the: P50, N100, and

P200; when the second stimulus was presented, in all groups. However, when the

auditory stimulus was followed by the visual stimulus, AD participants did not exhibit

a reduction in ERP amplitude during the refractory period, compared to HC and MCI.

HC participants experienced a reduced P50 component while MCI participants

experienced reduced P50 and N100 components. These results suggest that people

with AD have reduced cross-modal interactions compared to the HC and MCI groups.

Taking these studies into account, it is possible that people with MCI have an

increased TBW, compared to HC.

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

7

Methods

Participants

Fourteen older adults with mild cognitive impairment (10 males) between the

ages of 65 years and 78 years and sixteen age-matched healthy older adults (10 males)

took part in this experiment. Of the fourteen MCI participants, nine were diagnosed

as amnestic MCI and the remainder non-amnestic MCI. The participants with MCI

were recruited through the Goethe-University Memory Clinic. The MCI patients

came to the Memory Clinic, complaining of subjective memory loss or cognitive

decline, which was corroborated by a family member or close friend. The diagnose

criteria of MCI was based on an impairment of at least one of the following cognitive

domains: memory, language, visuospatial skills, executive function. This was

assessed using the Consortium to Establish a Registry for Alzheimer’s disease

(CERAD), with a cut-off criterial of at least one standard deviation below the age

norms. They were also given the Clinical Dementia Rating (CDR), with a score of

0.5.

None of the HC had a history of psychological disorder or report and sensory

impairment. The MCI population also did not report any auditory or visual

impairment. To ensure that the MCIs did not develop Alzheimer’s Disease between

the time of the initial diagnose and test, they were given a portion of the CERAD and

the D2 test to characterize their mental state (see Table 1 for a list of CERAD results).

The HCs were also given the CERAD and D2 to ensure that they did not have

undiagnosed MCI. There was no significant difference in the years of education

between the HCs (14.94 years) and MCI (15.7 years) (χ2 = 8.28, p < 0.51).

The ethics committee of the University of Frankfurt Medical Faculty approved

this experiment.

Apparatus and Stimuli

The visual stimuli were presented on a 24” flat panel computer monitor. The

visual stimulus was a white circular disk, subtending 2° of visual angle. This disk

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

8

was placed 8° of visual angle below the fixation cross. The presentation duration of

the disk was 16 ms.

The auditory stimulus consisted of a 16 ms, 3500 Hz pure tone with a total rise-

and decay-time of 20 µs at a sound pressure level of 68 dBA. They were presented

using closed, circum-aural headphones (AKG, Austria, model: K271) to reduce any

ambient noise.

Design and Procedures

The CERAD was given to each participant before behavioural testing. This was

done to ensure that the MCI participants did not develop Alzheimer’s disease, but also

to ensure that the older controls did not have undiagnosed MCI. The portions of the

CERAD given to participants were: semantic fluency (animals), Boston naming test,

mini-mental state, word learning, figure drawing and recall, phonemic fluency, and

the Trail Making Tests A and B (see Table 1 for a list of results). The norms of the

CERAD are adjusted for age, education level and sex.

The overall design of the experiment was based on a 3x7 Repeated Measures

Design with Modality (vision-only, auditory-only, and audiovisual) and Stimulus-

onset Asynchrony (SOA) as factors. The factor of Modality was blocked and the

order randomized between participants. At the beginning of each trial a fixation cross

was presented at the centre of the computer screen. Participants were instructed to

maintain their eye gaze on the cross throughout the experiment.

In the vision-only block, one or two flashes were presented and the participants’

task was to indicate how many flashes were presented. In the auditory-only block, one

or two beeps were presented and the participants indicated how many beeps they

heard. The SOA used in these blocks were: 0 ms (one stimulus event), 50 ms, 100

ms, 150 ms, 200 ms, 250 ms, 300 ms, 500 ms. In both blocks, a randomly permuted

SOA was used when two stimuli were presented. There were 130 trials in each of the

unimodal conditions, 60 trials where one stimulus was presented and the remaining

trials where two stimuli were presented, with an equal number divided between the

SOA conditions (10 trials in each remaining SOA condition).

The audio-visual block consisted of three conditions: two control conditions (1

beep/1 flash and 2 beeps/2 flashes) and the illusion condition (2 beeps/1 flash). In the

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

9

control conditions, the auditory and visual stimuli were presented simultaneously. In

the illusion condition, the visual flash was presented at the same time as the first

auditory beep. In the 2 beeps/2 flashes and 2 beep/1 flash conditions, the SOAs used

were 50 ms, 100 ms, 150 ms, 200 ms, 250 ms, 300 ms, 500 ms. These conditions were

randomly presented within the block to avoid any response biases. The participants’

task was to ignore the auditory stimuli and indicate how many visual flashes were

presented. The audio-visual block contained 210 trials, with 70 trials within each

condition.

All responses were made via a computer keyboard. It is possible that

participants could perceive more than the presented stimuli (audio or visual), thus;

responses were not restricted to “1” and “2”. In less than 1% of trials, participants

reported to have seen or heard three or more stimuli. While reaction times were

recorded participants were asked to emphasize accuracy over speed. The experiment

was programmed in Presentation (Neurobehavioral Systems, CA, USA).

Results

To examine the unimodal perception between the HC and MCI groups a

2x2x8 mixed-design ANOVA with Group (HC vs. MCI) as the between-subjects

factor, and Modality (vision-only vs. auditory-only) and SOA (0 ms, 50 ms, 100 ms,

150 ms, 200 ms, 250 ms, 300 ms, and 500 ms) as the within-subjects factors. All

analyses have been corrected for the lack of homogeneity of variance using the

Greenhouse-Geisser method. There was no main effect of Group [F(1,27) = 2.47,

partial eta-squared = 0.03, p = 0.13]. There was a main effect of Modality [F(1,27) =

12.52, partial eta-squared = 0.28, p = 0.001], with higher accuracy for the detection

of the auditory stimuli (90.37%) compared to the visual stimuli (81.86%). There was

also a main effect of SOA [F(7,189) = 53.29, partial eta-squared = 0.62, p < 0.0001].

Participants’ accuracy increased as the SOA between stimuli increased (0 ms =

91.29%; 50 ms = 53.84%; 100 ms = 80.85%; 150 ms = 89.18%; 200 ms = 93.30%;

250 ms = 94.06%; 300 ms = 93.93%; 500 ms = 92.46%). There was a significant

interaction between the factors of Modality and SOA [F(7,189) = 3.17, partial eta-

squared = 0.09, p = 0.034] (see Figure 1). A Newman-Keuls posthoc test revealed

participants were significantly worse when the SOA between the 2 flashes was 50 ms,

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

10

compared to all other SOAs (all p’s < 0.0005). There were no other significant

interactions.

To examine the reaction times during unimodal perception the same analysis

design was used. There was no main effect of Group [F(1,18) <1, n.s.]. There was

also no main effect of Modality [F(1,18,) < 1, n.s.]. Finally, there was no main effect

of SOA (F(7,126) = 1.6, p = 0.18]. There were no significant interactions between

these factors.

------------------ Place Figure 1 about here ------------------

To examine the audiovisual interactions between the two groups we used a

2x3x7 mixed-measures ANOVA with Group (HC vs. MCI) as the between-subjects

factor, and AV Stimuli (1-flash/1-beep vs. 2-flashes/2-beeps vs. 1-flash/2-beeps) and

SOA (50 ms, 100 ms, 150 ms, 200 ms, 250 ms, 300 ms, and 500 ms) as the within-

subjects factors. There was a main effect of Group [F(1, 33) = 9.55, partial eta-

squared = 0.22, p = 0.004]. Overall, HC performance (85.36%) was significantly

more accurate than MCI (76.28%). There was a main effect of AV Stimuli [F(2, 66)

= 76.81, partial eta-squared = 0.70, p < 0.0001]. Participants were significantly more

accurate in the control conditions (1 beep/1 flash = 93.74%; 2 beeps/2 flashes =

94.56%) compared to the illusion condition (54.15%). There was also a significant

main effect of SOA (50 ms = 71.73%; 100 ms = 73.42%; 150 ms = 61.61%; 200 ms =

83.56%; 250 ms = 82.94%; 200 ms = 83.71%; 500 ms = 88.77%) [F(6,198) = 10.24,

partial eta-squared = 0.24, p < 0.0001]. There was a significant interaction between

the factors of Group and AV Stimuli [F(2,66) = 5.213, partial eta-squared = 0.14, p =

0.026]. There was also a significant interaction between AV Stimuli and SOA

[F(12,396) = 8.98, partial eta-squared = 0.21, p < 0.0001]. There was no significant

three-way interaction.

Using a Mann-Whitney non-parametric test between the two groups, there was

a significant difference between the HC and MCI groups in the 1 beep/1 flash 200 ms

(Z = -2.92, p = 0.003) and 1 beep/1 flash 300 ms (Z = -2.56, p = 0.01) conditions,

with MCI participants being significantly worse than HCs. Importantly, MCI

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

11

participants perceived significantly more illusions in the 2 beeps/1flash condition

when the SOA between the two beeps was 150 ms (Z = -3.38, p = 0.0007), 250 ms (Z

= -2.60, p = 0.009), and 300 ms (Z = -2.58, p = 0.01) (see Figure 2A).

To examine the reaction times during multisensory perception the same

analysis design was used. There was no main effect of Group [F(1,11) < 1, n.s.].

There was a main effect of AV Stimuli [F(2,22) = 5.00, p = 0.027]. Participants were

significantly slower in the 2 beeps/1 flash (illusion) condition (1246.22 ms) compared

to the 2 beeps/2 flashes (1136.91 ms) and 1 beep/1 flash conditions (1074.20 ms).

There was also a significant difference between these two AV control conditions (all

ps < 0.05). There was also a significant main effect of SOA (50 ms = 1237.40 ms;

100 ms = 1214.76 ms; 150 ms = 1164.58 ms; 200 ms = 1133.08 ms; 250 ms =

1103.30 ms; 300 ms = 1138 ms; 500 ms = 1075.91 ms) [F(6,66) = 6.078, p = 0.001].

Participants responded more quickly as the SOA between the two stimuli increased.

There was a significant interaction between AV Stimuli and SOA. Participants were

significantly faster in the 2 beeps/1 flash condition when the SOA between two beeps

was 500 ms, compared for the other SOAs (all ps < 0.001). There was no significant

difference between the SOAs in the 1 flash/1 beeps and 2 flashes/2 beeps conditions.

There was a significant interaction between AV Stimuli and Group [F(2,22) = 9.99, p

= 0.003]. The MCI participants (1388.06 ms) were significantly slower in the 2

beeps/1 flash condition, compared to the HC (1076.47 ms). There were significant

differences between the two groups in the 1 beep/1flash and 2 beeps/2 flashes

conditions (see Figure 2b). Although, there was a trend in the 1 beep/1 flash

condition for longer reaction times for the MCI participants (1125.76 ms) compared

to the HC (976.14 ms; p = 0.052). There was no significant interaction between SOA

and Group [F(6,66) = 1.79, p = 0.160]. Finally, there was no significant three-way

interaction between these factors.

A Mann-Whitney non-parametric test between the two groups revealed that

MCIs were significantly slower than HCs in the 1 beep/1 flash condition when the

SOA was 200 ms (Z = -2.92, P = 0.003) and 300 ms (Z = -2.56, p = 0.01). In the 2

beep/1 flash (illusion) condition MCIs were significantly slower than HCs when the

SOA between the two beeps were 150 ms (Z = -3.38, p < 0.0005), 250 ms (Z = -2.6, p

< 0.01), and 300 ms (Z = -2.58, p < 0.01).

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

12

------------------ Place Figure 2 about here ------------------

Discussion

MCI is only characterized by a person’s performance in cognitive tasks.

However these data, as well as previous studies [32], have demonstrated that people

with MCI also show behavioural differences in early sensory integration processing,

compared to HC. These findings revealed, participants with MCI perceived more SiFi

compared to healthy controls. MCIs perceived more illusions, across a broader range

of auditory SOAs compared to HC. This suggests that not only are MCIs more likely

to integrate audio-visual information, compared to controls; MCIs also integrate

audio-visual information across a ‘wider’ temporal binding window than controls.

There was no significant difference between the two groups in the 2 beeps/1 flash

condition when the SOA between the auditory stimuli was 500 ms; suggesting that the

results are not due to a difference in understanding the task. These results suggest that

this SiFi behavioural task is sensitive to delineate between healthy older adults and

those older adults with mild cognitive impairment.

In order to determine whether the behavioural differences in the SiFi are

associated with a specific cognitive deficit we performed a correlational analysis

between the different sections of the CERAD compared to their individual

behavioural performance. We did not find any significant correlation between the

SiFi and any subsection of the CERAD in the MCI and older adults groups. It

appears that the SiFi is not related to a particular cognitive decline but to the overall

deficit. Perhaps this is not surprising, given that the SiFi is known to be an early

sensory process [38].

It is possible that the difference in the number of perceived illusions between

the groups is related to differences in unisensory acuity. This may result in a

modulation of the perceived reliability (maximum likelihood estimate) between the

auditory and visual signals. For example, if participants were not able to accurately

determine the number of perceived visual flashes, but accurately determine the

number of beeps, participants will rely more on the auditory modality than the visual

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

13

modality. However, there was no significant difference between the groups in the two

unimodal conditions.

The sound-induced flash illusion has been used to find differences in

multisensory integration in healthy older adults compared to young adults, healthy

older adults compared to older adults prone to falling, children with ASD compared to

healthy age-matched controls, and now people with MCI compared to healthy

controls. This relatively simple behavioural paradigm is a powerful tool that can

potentially be used to assess the cortical connectivity between sensory regions.

According to Keil et al. (2013), the SiFi is regulated by pre-stimulus beta-band

activity, in young adults. Beta-band activity is thought of as a correlate of top-down

information transfer [49-51]. Using phase-locking values, Keil et al. found increased

connectivity for trials in which the illusions were perceived, between the areas of the

left middle temporal gyrus and the primary visual and auditory areas. EEG beta-band

activity in the occipital lobe has been implicated in increased arousal and visual

attention [see 49 for a review; 50; 52]. However, people with MCI do not exhibit

such significant changes in beta-band activity [53]. Beta-band activity is also critical

for the integration of audio-visual stimuli [54]. Given that the SiFi is modulated by

pre-stimulus beta-band activity [39], it is possible that the participants with MCI had

less modulation of their inter-trial beta-band activity, resulting in the perception of

more illusions, compared to HCs.

It has been demonstrated that healthy older adults with decreased unisensory

perception also have increased cognitive decline [1; 55]. It could be that the cognitive

decline is simply the result of increased sensory noise. Baltes and Lindenberger [1]

suggested that cognitive decline is not the result of increasingly noisy signals but due

to a breakdown of a cortical network that is common to both sensory perception and

fluid intelligence; however, they do not propose candidate brain regions. More

recently, Melnick, Harrison, Park, Bennetto, and Tadin [56] behaviourally

demonstrated that the link between sensory acuity and fluid intelligence [see 57; 58

for a link between fluid intelligence and working memory] may not be due to a

common neural framework, but correlated with an individual’s overall ability inhibit

surrounding (irrelevant) information.

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

14

It is possible that the perceptual differences are due to differences in

‘processing speed’ [59; 60]. Deary et al. [60] have demonstrated that visual

information processing is correlated with general intelligence. However, it is unlikely

that processing speed is the sole reason for MCIs perceiving more illusions than HCs.

Melnick et al. [56] demonstrated in young healthy controls that processing speed (and

intelligence) cannot entirely account for visual perception in a discrimination task.

Visual perception performance was also correlated with poor stimulus inhibition.

While MCIs can have high general intelligence, it is also possible that they exhibit a

reduced ability to suppress irrelevant information.

Our results extend the findings of Melnick et al. by suggesting that it is the

MCIs’ inability to supress irrelevant auditory stimuli results in the increased

perception of illusory flashes. There are common brain areas used in both the

processing of working memory and the SiFi. Increased beta-band power in the left

middle frontal gyus (BA9) as well as BA4 of the parietal cortex has been associated

with SiFi perception [39] and working memory [as well as bilateral dorsolateral

prefrontal cortex and inferior parietal cortex-BA 40/7; 58]. This does not suggest that

there is a common network for both the SiFi and working memory but there are some

common brain areas involved in both tasks.

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

15

References

[1]BaltesPB,LindenbergerU.Emergenceofapowerfulconnectionbetweensensory

andcognitivefunctionsacrosstheadultlifespan:anewwindowtothestudyof

cognitiveaging?PsychologyandAging12(1):12‐21.(1997).

[2]ShamsL,KamitaniY,ShimojoS.Whatyouseeiswhatyouhear.Nature408:

788.(2000).

[3]ShamsL,KamitaniY,ShimojoS.Visualillusioninducedbysound.CognitiveBrain

Research14:147‐52.(2002).

[4]RitchieK,LovestoneS.Thedementias.TheLancet360(9347):1759‐66.(2002).

[5]RitchieK,TouchonJ.HeterogeneityinseniledementiaoftheAlzheimertype:

Individualdifferences,progressivedeteriorationorclinicalsub‐types?JournalofClinical

Epidemiology45(12):1391‐98.(1992).

[6]HoganMJ.Thecerebelluminthoughtandaction:afronto‐cerebellaraging

hypothesis.NewIdeasinPsychology22(2):97‐125.(2004).

[7]HoganMJ,SwanwickGRJ,KaiserJ,RowanM,LawlorB.Memory‐relatedEEGpower

andcoherencereductionsinmildAlzheimer'sdisease.InternationalJournalof

Psychophysiology49(2):147‐63.(2003).

[8]GradyCL,CraikFIM.Changesinmemoryprocessingwithage.CurrentOpinionin

Neurobiology10:224‐31.(2000).

[9]AndersonND,CraikFIM.Memoryintheageingbrain.In:TheOxfordhandbookof

memory.(Eds:TulvingE,CraikFIM).Oxford,UK:OxfordUniversityPress;pp.411‐25

(2000).

[10]LiS‐C,LindenbergerU.Cross‐levelunification:Acomputationalexplorationofthe

linkbetweendeteriorationofneurotransmittersystemsanddedifferentiationof

cognitiveabilitiesinoldage.In:Cognitiveneuroscienceofmemory.(Eds:NilssonLG,

MarkowitschH).Seattle,WA,USA:Hogrefe&Huber;pp.103‐46(1999).

[11]HoganMJ,KellyCAM,CraikFIM.TheEffectsofAttentionSwitchingonEncoding

andRetrievalofWordsinYoungerandOlderAdults.ExperimentalAgingResearch

32(2):153‐83.(2006).

[12]SalthouseTA.Agingandmeasuresofprocessingspeed.BiologicalPsychology54(1–

3):35‐54.(2000).

[13]LindenbergerU,MarsiskeM,BaltesPB.Memorizingwhilewalking:Increasein

dual‐Taskcostsfromyoungadulthoodtooldage.PsychologyandAging15(3):417‐

36.(2000).

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

16

[14]FreiherrJ,LundströmJN,HabelU,ReetzK.Multisensoryintegrationmechanisms

duringaging.FrontiersinHumanNeuroscience72013).

[15]StaalJA,PinkneyL,RoaneDM.AssessmentofStimulusPreferencesinMultisensory

EnvironmentTherapyforOlderPeoplewithDementia.TheBritishJournalof

OccupationalTherapy66(12):542‐50.(2003).

[16]FeldmanH,ScheltensP,ScarpiniE,HermannN,MesenbrinkP,MancioneL,etal.

Behavioralsymptomsinmildcognitiveimpairment.Neurology62(7):1199‐201.(2004).

[17]GauthierS,ReisbergB,ZaudigM,PetersenRC,RitchieK,BroichK,etal.Mild

cognitiveimpairment.TheLancet367(9518):1262‐70.(2006).

[18]BischkopfJ,BusseA,AngermeyerMC.Mildcognitiveimpairment–areviewof

prevalence,incidenceandoutcomeaccordingtocurrentapproaches.ActaPsychiatr

Scand106:403‐14.(2002).

[19]RitchieK,TouchonJ.Mildcognitiveimpairment:conceptualbasisandcurrent

nosologicalstatus.TheLancet355(9199):225‐28.(2000).

[20]BurnsAS,ZaudigM.Mildcognitiveimpairmentinolderpeople.TheLancet360:

1963–65.(2002).

[21]PetersenRC.Mildcognitiveimpairmentasadiagnosticentity.JournalofInternal

Medicine256:183‐94.(2004).

[22]PetersenRC,RobertsRO,KnopmanDS,BoeveBF,GedaYE,IvnikRJ,etal.Mild

CognitiveImpairment:TenYearsLater.ArchivesofNeurology66(12):1447‐55.(2009).

[23]LuckT,LuppaM,BrielS,Riedel‐HellerSG.IncidenceofMildCognitiveImpairment:

ASystematicReview.DementiaandGeriatricCognitiveDisorders29(2):164‐75.(2010).

[24]BokdeALW,Lopez‐BayoP,MeindlT,PechlerS,BornC,FaltracoF,etal.Functional

connectivityofthefusiformgyrusduringaface‐matchingtaskinsubjectswithmild

cognitiveimpairment.Brain129(5):1113‐24.(2006).

[25]BokdeALW,Lopez‐BayoP,BornC,DongW,MeindlT,LeinsingerG,etal.Functional

abnormalitiesofthevisualprocessingsysteminsubjectswithmildcognitive

impairment:AnfMRIstudy.PsychiatryResearch:Neuroimaging163(3):248‐59.(2008).

[26]WenigerG,RuhlederM,LangeC,WolfS,IrleE.Egocentricandallocentricmemory

asassessedbyvirtualrealityinindividualswithamnesticmildcognitiveimpairment.

Neuropsychologia49(3):518‐27.(2011).

[27]Alescio‐LautierB,MichelBF,HerreraC,ElahmadiA,ChambonC,TouzetC,etal.

Visualandvisuospatialshort‐termmemoryinmildcognitiveimpairmentandAlzheimer

disease:Roleofattention.Neuropsychologia45(8):1948‐60.(2007).

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

17

[28]BushG,WhalenPJ,RosenBR,JenikeMA,McInerneySC,RauchSL.Thecounting

Stroop:aninterferencetaskspecializedforfunctionalneuroimaging‐‐validationstudy

withfunctionalMRI.HumanBrainMapping6(4):270‐82.(1998).

[29]RahmanTTA,MohamedST,AlbanoubyMH,BekhetHF.Centralauditoryprocessing

inelderlywithmildcognitiveimpairment.Geriatrics&GerontologyInternational:no‐

no.(2011).

[30]IdrizbegovicE,HederstiernaC,DahlquistM,KämpfeNordströmC,JelicV,Rosenhall

U.CentralauditoryfunctioninearlyAlzheimer'sdiseaseandinmildcognitive

impairment.AgeandAgeing40(2):249‐54.(2011).

[31]MapstoneM,SteffenellaTM,DuffyCJ.Avisuospatialvariantofmildcognitive

impairment:GettinglostbetweenagingandAD.Neurology60(5):802‐08.(2003).

[32]WuJ,YangJ,YuY,LiQ,NakamuraN,ShenY,etal.DelayedAudiovisualIntegration

ofPatientswithMildCognitiveImpairmentandAlzheimer'sDiseaseComparedwith

NormalAgedControls.JournalofAlzheimer'sDisease32(2):317‐28.(2012).

[33]SettiA,BurkeKE,KennyRA,NewellFN.Isinefficientmultisensoryprocessing

associatedwithfallsinolderpeople?ExperimentalBrainResearch2209:375‐

284.(2011).

[34]PoliakoffE,ShoreDI,LoweC,SpenceC.Visuotactiletemporalorderjudgmentsin

ageing.NeuroscienceLetters396(3):207‐11.(2006).

[35]LaurientiPJ,BurdetteJH,MaldjianJA,WallaceMT.Enhancedmultisensory

integrationinolderadults.NeurobiologyofAging27(8):1155‐63.(2006).

[36]SettiA,FinniganS,SobolewskiR,McLarenL,RobertsonIH,ReillyRB,etal.

Audiovisualtemporaldiscriminationislessefficientwithaging:Anevent‐related

potentialstudy.Neuroreport22(11):554‐58.(2011).

[37]DeLossDJ,PierceRS,AndersenGJ.MultisensoryIntegration,Aging,andtheSound‐

InducedFlashIllusion.PsychologyandAging2013).

[38]MishraJ,MartinezA,SejnowskiTJ,HillyardSA.EarlyCross‐ModalInteractionsin

AuditoryandVisualCortexUnderlieaSound‐InducedVisualIllusion.Journalof

Neuroscience27(15):4120‐31.(2007).

[39]KeilJ,MüllerN,HartmannT,WeiszN.PrestimulusBetaPowerandPhase

SynchronyInfluencetheSound‐InducedFlashIllusion.CerebralCortex:1‐7.(2013).

[40]RoopunAK,MiddletonSJ,CunninghamMO,LeBeauFEN,BibbigA,WhittingtonMA,

etal.Abeta2‐frequency(20–30Hz)oscillationinnonsynapticnetworksof

somatosensorycortex.ProceedingsoftheNationalAcademyofSciences103(42):

15646‐50.(2006).

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

18

[41]DuerdenEG,TaylorMJ,SooryaLV,WangT,FanJ,AnagnostouE.Neuralcorrelates

ofinhibitionofsociallyrelevantstimuliinadultswithautismspectrumdisorder.Brain

Research1533(0):80‐90.(2013).

[42]LeuchterAF,NewtonTF,CookIA,O.WD,Rosenberg‐ThompsonS,LachenbruchPA.

ChangesinbrainfunctionalconnectivityinAlzheimer‐typeandmulti‐infarctdementia.

Brain115:1543‐61.(1992).

[43]DelbeuckX,VanderLindenM,ColletteF.Alzheimer'DiseaseasaDisconnection

Syndrome?NeuropsychologyReview13(2):79‐92.(2003).

[44]Gómez‐IslaT,HollisterR,WestH,MuiS,GrowdonJH,PetersenRC,etal.Neuronal

losscorrelateswithbutexceedsneurofibrillarytanglesinAlzheimer'sdisease.Annalsof

Neurology41(1):17‐24.(1997).

[45]HoganMJ,StaffRT,BuntingBP,MurrayAD,AhearnTS,DearyIJ,etal.Cerebellar

brainvolumeaccountsforvarianceincognitiveperformanceinolderadults.Cortex

47(4):441‐50.(2011).

[46]DelbeuckX,ColletteF,VanderLindenM.IsAlzheimer'sdiseaseadisconnection

syndrome?:Evidencefromacrossmodalaudio‐visualillusoryexperiment.

Neuropsychologia45(14):3315‐23.(2007).

[47]GolobEJ,MirandaGG,JohnsonJK,StarrA.Sensorycorticalinteractionsinaging,

mildcognitiveimpairment,andAlzheimer'sdisease.NeurobiologyofAging22(5):755‐

63.(2001).

[48]NäätänenR,PictonT.TheN1WaveoftheHumanElectricandMagneticResponse

toSound:AReviewandanAnalysisoftheComponentStructure.Psychophysiology

24(4):375‐425.(1987).

[49]EngelAK,FriesP.Beta‐bandoscillations‐signallingthestatusquo?Current

OpinioninNeurobiology20(2):156‐65.(2010).

[50]LeeJH,WhittingtonMA,KopellNJ.Top‐DownBetaRhythmsSupportSelective

AttentionviaInterlaminarInteraction:AModel.PLoSComputBiol9(8):

e1003164.(2013).

[51]SalehM,ReimerJ,PennR,OjakangasCL,HatsopoulosNG.FastandSlow

OscillationsinHumanPrimaryMotorCortexPredictOncomingBehaviorallyRelevant

Cues.Neuron65(4):461‐71.(2010).

[52]WróbelA.Betaactivity:acarrierforvisualattention.ActaNeurobiol60:247‐

60.(2000).

[53]GüntekinB,Emek‐SavaşDD,KurtP,YenerGG,BaşarE.Betaoscillatoryresponsesin

healthysubjectsandsubjectswithmildcognitiveimpairment.NeuroImage:Clinical

3(0):39‐46.(2013).

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

19

[54]SenkowskiD,MolholmS,Gomez‐RamirezM,FoxeJJ.OscillatoryBetaActivity

PredictsResponseSpeedduringaMultisensoryAudiovisualReactionTimeTask:A

High‐DensityElectricalMappingStudy.CerebralCortex16(11):1556‐65.(2006).

[55]LindenbergerU,BaltesPB.Sensoryfunctionalandintelligenceinoldage:Astrong

connection.PsychologyandAging9(3):339‐55.(1994).

[56]MelnickMichaelD,HarrisonBryanR,ParkS,BennettoL,TadinD.AStrong

InteractiveLinkbetweenSensoryDiscriminationsandIntelligence.CurrentBiology

23(11):1013‐17.(2013).

[57]BurgessGC,BraverTS.NeuralMechanismsofInterferenceControlinWorking

Memory:EffectsofInterferenceExpectancyandFluidIntelligence.PLoSONE5(9):

e12861.(2010).

[58]BurgessGC,GrayJR,ConwayARA,BraverTS.Neuralmechanismsofinterference

controlunderlietherelationshipbetweenfluidintelligenceandworkingmemoryspan.

JournalofExperimentalPsychology:General140(4):674‐92.(2011).

[59]DearyIJ.Intelligence.AnnualReviewofPsychology63(1):453‐82.(2012).

[60]DearyIJ,McCrimmonRJ,BradshawJ.Visualinformationprocessingand

intelligence.Intelligence24(3):461‐79.(1997).

Acknowledgements

This study was supported by the Hessian initiative for the development of scientific

and economic excellence (LOEWE) Neuronal Coordination Research Focus Frankfurt

(NeFF).

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

20

Table 1.

Table 1: The mean CERAD results from each group. The values represent the

average score for each section. The numbers in parentheses represent the standard

error of the mean.

CERAD results

HC MCI

Semantic Fluency (animals) 23.69 (0.69) 15.25 (1.12)

Boston Naming Test 14.54 (0.89) 13.5 (0.87)

Mini- Mental State 29.23 (0.25) 27.5 (1.2)

Verbal Learning Immediate Recall (total) 22.38 (0.75) 15.75 (1.39)

Delayed Recall 8.23 (0.64) 5.5 (1.15)

Constructional Praxis Drawing 10.54 (1.02) 10.75 (0.55)

Replicate drawings 10.08 (0.99) 9 (1.07)

Savings 0.96 (1.01) 0.83 (0.87)

Word Fluency 18.23 (0.51) 9.25 (1.08)

Trail Making Test Part A 35.46 (1.14) 40.5 (0.99)

Part B 84.54 (1.33) 163.75 (1.33)    Part B/A 2.52 (1.26) 4.2 (1.09)

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

21

Figure Captions.

1) Results from the visual-only and auditory-only blocks. Both groups were

more accurate when determining the number of beeps, compared to flashes. In

both modalities, participants were more accurate as the SOA between stimuli

increased. There was no significant difference between groups. The error

bars represent the SEM.

2) A) This graph illustrates the mean accuracy from the multisensory conditions.

There were no group differences in the control conditions (1 beep/1 flash & 2

beeps/2 flashes). MCI participants perceived more illusions, across a wider

range of SOAs, compared to HC. The error bars represent the SEM. B) The

mean reaction times in the multisensory conditions. MCI participants took

significantly longer in the 2 beeps/1 flash condition, compared to HCs.

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

22

0

10

20

30

40

50

60

70

80

90

100

0ms 50ms 100ms 150ms 200ms 250ms 300ms 500ms

1stimulus 2stimuli

Accuracy

Condition

UnimodalConditions

Vision

Auditory

Figure 1.

EXPANDINGTEMPORALBINDINGWINDOWSINMCI

23

0

10

20

30

40

50

60

70

80

90

100

50ms 100ms 150ms 200ms 250ms 300ms 500ms

Accuracy

SOA

Accuracy

HC:1beep/1flash

HC:2beep/2flash

HC:2beeps/1flash

MCI:1beep/1flash

MCI:2beep/2flash

MCI:2beeps/1flash

0

200

400

600

800

1000

1200

1400

1600

1beep/1flash 2beep/2flash 2beeps/1flash

MeanReactionTimes(ms)

AVStimuli

ReactionTime

HC

MCI

Figure 2.

A)

B)